商城类网站怎么优化,关键词优化包年推广,阿里企业邮箱申请,百姓网招聘信息首先是环境#xff1a; 我是在Anaconda3中的Jupyter Notebook (tensorflow)中进行训练#xff0c;环境各位自行安装 数据集#xff1a; 本次数据集五个类型#xff08;键盘#xff0c;椅子#xff0c;眼镜#xff0c;水杯#xff0c;鼠标#xff09;我收集了每个接近两…首先是环境 我是在Anaconda3中的Jupyter Notebook (tensorflow)中进行训练环境各位自行安装 数据集 本次数据集五个类型键盘椅子眼镜水杯鼠标我收集了每个接近两千张的图片共11091张 这个可以不用这么多因为cnn模型训练也用不上这么多的图片可以自行减少这个是我这边的要求所以我索性就直接训练了。 预测结果如下 代码如下 相关库
import os
import numpy as np
from PIL import Image
from sklearn.model_selection import train_test_split
import matplotlib.pyplot as plt
import tensorflow as tf
from tensorflow.keras import models, layers
图片数据处理
# 首先导入必要的库并设置PIL的限制
import os
import numpy as np
from PIL import Image
Image.MAX_IMAGE_PIXELS None # 解除PIL的图片大小限制
from sklearn.model_selection import train_test_split
import matplotlib.pyplot as plt
import tensorflow as tf
from tensorflow.keras import models, layers
import warnings
warnings.filterwarnings(ignore) # 忽略警告信息def load_and_preprocess_data(base_path, img_size(300, 300), batch_size32):# 初始化列表存储图片路径和标签image_paths []labels []# 类别映射class_mapping {jp: 0, # 键盘yz: 1, # 椅子yj: 2, # 眼镜bz: 3, # 水杯sb: 4 # 鼠标}# 首先检查并收集有效的图片路径和标签print(正在检查图片文件...)for filename in os.listdir(base_path):if filename.endswith((.jpg, .png)):prefix filename[:2]if prefix in class_mapping:try:img_path os.path.join(base_path, filename)# 尝试打开图片验证其有效性with Image.open(img_path) as img:image_paths.append(img_path)labels.append(class_mapping[prefix])print(f成功验证图片: {filename})except Exception as e:print(f跳过无效图片 {filename}: {str(e)})continueif not image_paths:raise ValueError(没有找到有效的图片文件)# 转换标签为numpy数组labels np.array(labels)# 创建数据生成器class ImageDataGenerator:def __init__(self, image_paths, labels, img_size, batch_size):self.image_paths image_pathsself.labels labelsself.img_size img_sizeself.batch_size batch_sizeself.n len(image_paths)self.indexes np.arange(self.n)np.random.shuffle(self.indexes)self.i 0def __len__(self):return (self.n self.batch_size - 1) // self.batch_sizedef __iter__(self):return selfdef __next__(self):if self.i self.n:self.i 0np.random.shuffle(self.indexes)raise StopIterationbatch_indexes self.indexes[self.i:min(self.i self.batch_size, self.n)]batch_paths [self.image_paths[i] for i in batch_indexes]batch_labels self.labels[batch_indexes]batch_images []valid_labels []for path, label in zip(batch_paths, batch_labels):try:with Image.open(path) as img:# 转换为RGB模式if img.mode ! RGB:img img.convert(RGB)# 调整图片大小if img.size[0] 1000 or img.size[1] 1000:img.thumbnail((1000, 1000), Image.Resampling.LANCZOS)img img.resize(self.img_size, Image.Resampling.LANCZOS)# 转换为numpy数组img_array np.array(img, dtypenp.float32) / 255.0batch_images.append(img_array)valid_labels.append(label)except Exception as e:print(f处理图片 {path} 时出错: {str(e)})continueself.i self.batch_sizeif not batch_images: # 如果这个批次没有有效图片return self.__next__() # 尝试下一个批次return np.array(batch_images), np.array(valid_labels)# 打印数据集信息print(f\n总共找到 {len(image_paths)} 张有效图片)for label in set(labels):count np.sum(labels label)print(f类别 {label}: {count} 张图片)# 划分训练集和测试集的索引n_samples len(image_paths)n_train int(0.8 * n_samples)indices np.random.permutation(n_samples)train_idx, test_idx indices[:n_train], indices[n_train:]# 创建训练集和测试集的生成器train_generator ImageDataGenerator([image_paths[i] for i in train_idx],labels[train_idx],img_size,batch_size)test_generator ImageDataGenerator([image_paths[i] for i in test_idx],labels[test_idx],img_size,batch_size)return train_generator, test_generator# 修改训练函数中的训练循环
def train_and_save_model(model, train_generator, test_generator, model_name, epochs10):# 编译模型model.compile(optimizeradam,losssparse_categorical_crossentropy,metrics[accuracy])# 训练模型history {accuracy: [], val_accuracy: [], loss: [], val_loss: []}for epoch in range(epochs):print(f\nEpoch {epoch1}/{epochs})# 训练阶段train_loss []train_acc []for i, (batch_images, batch_labels) in enumerate(train_generator):if len(batch_images) 0:continuemetrics model.train_on_batch(batch_images, batch_labels)train_loss.append(metrics[0])train_acc.append(metrics[1])print(f\rBatch {i1} - loss: {metrics[0]:.4f} - accuracy: {metrics[1]:.4f}, end)# 验证阶段val_loss []val_acc []for batch_images, batch_labels in test_generator:if len(batch_images) 0:continuemetrics model.test_on_batch(batch_images, batch_labels)val_loss.append(metrics[0])val_acc.append(metrics[1])# 记录历史epoch_train_loss np.mean(train_loss) if train_loss else 0epoch_train_acc np.mean(train_acc) if train_acc else 0epoch_val_loss np.mean(val_loss) if val_loss else 0epoch_val_acc np.mean(val_acc) if val_acc else 0history[accuracy].append(epoch_train_acc)history[val_accuracy].append(epoch_val_acc)history[loss].append(epoch_train_loss)history[val_loss].append(epoch_val_loss)print(f\nEpoch {epoch1} - loss: {epoch_train_loss:.4f} - accuracy: {epoch_train_acc:.4f} - fval_loss: {epoch_val_loss:.4f} - val_accuracy: {epoch_val_acc:.4f})# 绘制训练历史plt.figure(figsize(12, 4))plt.subplot(1, 2, 1)plt.plot(history[accuracy], label训练准确率)plt.plot(history[val_accuracy], label验证准确率)plt.title(模型准确率)plt.xlabel(Epoch)plt.ylabel(准确率)plt.legend()plt.subplot(1, 2, 2)plt.plot(history[loss], label训练损失)plt.plot(history[val_loss], label验证损失)plt.title(模型损失)plt.xlabel(Epoch)plt.ylabel(损失)plt.legend()plt.savefig(f{model_name}_training_history.png)# 保存模型model.save(f{model_name}.h5)
模型训练
#这里的epochs的数值为2代表训练2次各位可以自行更改
def train_and_save_model(model, train_generator, test_generator, model_name, epochs2):# 编译模型model.compile(optimizeradam,losssparse_categorical_crossentropy,metrics[accuracy])# 训练模型history {accuracy: [], val_accuracy: [], loss: [], val_loss: []}for epoch in range(epochs):print(f\nEpoch {epoch1}/{epochs})# 训练阶段train_loss []train_acc []for batch_images, batch_labels in train_generator:metrics model.train_on_batch(batch_images, batch_labels)train_loss.append(metrics[0])train_acc.append(metrics[1])# 验证阶段val_loss []val_acc []for batch_images, batch_labels in test_generator:metrics model.test_on_batch(batch_images, batch_labels)val_loss.append(metrics[0])val_acc.append(metrics[1])# 记录历史history[accuracy].append(np.mean(train_acc))history[val_accuracy].append(np.mean(val_acc))history[loss].append(np.mean(train_loss))history[val_loss].append(np.mean(val_loss))print(floss: {np.mean(train_loss):.4f} - accuracy: {np.mean(train_acc):.4f} - fval_loss: {np.mean(val_loss):.4f} - val_accuracy: {np.mean(val_acc):.4f})# 绘制训练历史plt.figure(figsize(12, 4))plt.subplot(1, 2, 1)plt.plot(history[accuracy], label训练准确率)plt.plot(history[val_accuracy], label验证准确率)plt.title(模型准确率)plt.xlabel(Epoch)plt.ylabel(准确率)plt.legend()plt.subplot(1, 2, 2)plt.plot(history[loss], label训练损失)plt.plot(history[val_loss], label验证损失)plt.title(模型损失)plt.xlabel(Epoch)plt.ylabel(损失)plt.legend()plt.savefig(f{model_name}_training_history.png)# 保存模型model.save(f{model_name}.h5) 主程序
# 设置数据集路径
base_path E:/modoltry:# 检查路径是否存在if not os.path.exists(base_path):raise FileNotFoundError(f找不到指定路径{base_path})# 数据预处理print(正在加载和预处理数据...)train_generator, test_generator load_and_preprocess_data(base_path, batch_size32)# 训练CNN模型print(\n正在训练CNN模型...)cnn_model create_cnn_model()train_and_save_model(cnn_model, train_generator, test_generator, cnn_model)print(\n模型训练完成可以运行GUI程序进行测试。)except Exception as e:print(f\n程序出错{str(e)})
然后是GUI界面
#单cnn模型gui界面
import tkinter as tk
from tkinter import filedialog, ttk
from PIL import Image, ImageTk
import numpy as np
import tensorflow as tfclass ObjectClassifierGUI:def __init__(self, root):self.root rootself.root.title(物体分类器)self.root.geometry(800x600) # 设置窗口大小# 加载模型self.model tf.keras.models.load_model(cnn_model.h5)# 类别标签self.classes [键盘, 椅子, 眼镜, 水杯, 鼠标]# 创建GUI组件self.create_widgets()def create_widgets(self):# 创建主框架main_frame ttk.Frame(self.root, padding10)main_frame.grid(row0, column0, sticky(tk.W, tk.E, tk.N, tk.S))# 创建按钮框架button_frame ttk.Frame(main_frame)button_frame.grid(row0, column0, columnspan2, pady10)# 选择图片按钮self.select_btn ttk.Button(button_frame, text选择图片, commandself.select_image)self.select_btn.pack(sidetk.LEFT, padx5)# 清除图片按钮self.clear_btn ttk.Button(button_frame,text清除图片,commandself.clear_image,statedisabled # 初始状态为禁用)self.clear_btn.pack(sidetk.LEFT, padx5)# 创建左右分栏left_frame ttk.Frame(main_frame)left_frame.grid(row1, column0, padx10)right_frame ttk.Frame(main_frame)right_frame.grid(row1, column1, padx10)# 图片显示区域左侧ttk.Label(left_frame, text选择的图片).pack(pady5)self.image_label ttk.Label(left_frame)self.image_label.pack(pady5)# 预测结果显示右侧ttk.Label(right_frame, text预测结果).pack(pady5)self.result_frame ttk.Frame(right_frame)self.result_frame.pack(pady5)# 预测结果详细信息self.pred_class_label ttk.Label(self.result_frame,text预测类别-,font(Arial, 12))self.pred_class_label.pack(pady5)self.confidence_label ttk.Label(self.result_frame,text置信度-,font(Arial, 12))self.confidence_label.pack(pady5)# 所有类别的概率分布self.prob_frame ttk.Frame(self.result_frame)self.prob_frame.pack(pady10)self.prob_bars []for i in range(len(self.classes)):ttk.Label(self.prob_frame, textf{self.classes[i]}:).grid(rowi, column0, padx5)prob_bar ttk.Progressbar(self.prob_frame, length200, modedeterminate)prob_bar.grid(rowi, column1, padx5)prob_value ttk.Label(self.prob_frame, text0%)prob_value.grid(rowi, column2, padx5)self.prob_bars.append((prob_bar, prob_value))def select_image(self):# 打开文件选择对话框file_path filedialog.askopenfilename()if file_path:# 处理并显示图片image Image.open(file_path)# 保持原始宽高比例缩放图片用于显示display_size (300, 300)image.thumbnail(display_size, Image.Resampling.LANCZOS)# 显示图片photo ImageTk.PhotoImage(image)self.image_label.configure(imagephoto)self.image_label.image photo# 预处理图片用于预测image_for_pred image.resize((300, 300))img_array np.array(image_for_pred) / 255.0self.current_image_array np.expand_dims(img_array, axis0)# 进行预测self.predict_image(self.current_image_array)# 启用清除按钮self.clear_btn[state] normaldef predict_image(self, img_array):# 使用模型预测predictions self.model.predict(img_array)# 获取预测结果pred_class_idx np.argmax(predictions[0])pred_class self.classes[pred_class_idx]confidence predictions[0][pred_class_idx] * 100# 更新预测类别和置信度self.pred_class_label.config(textf预测类别{pred_class})self.confidence_label.config(textf置信度{confidence:.2f}%)# 更新所有类别的概率条for i, ((bar, value_label), prob) in enumerate(zip(self.prob_bars, predictions[0])):percentage prob * 100bar[value] percentagevalue_label.config(textf{percentage:.1f}%)def clear_image(self):# 清除图片显示self.image_label.configure(image)self.image_label.image None# 重置预测结果self.pred_class_label.config(text预测类别-)self.confidence_label.config(text置信度-)# 重置概率条for bar, value_label in self.prob_bars:bar[value] 0value_label.config(text0%)# 禁用清除按钮self.clear_btn[state] disabled# 清除存储的图像数组if hasattr(self, current_image_array):del self.current_image_array# 主程序
if __name__ __main__:root tk.Tk()app ObjectClassifierGUI(root)root.mainloop() 另外是一个调取摄像头实时识别的页面但是这个精度不是很高可能是摄像头的画面太杂了就不分享了。
